Search

Numerous conceptual frameworks exist for good practices in research data and analysis (e.g. Open Science and FAIR principles). In practice, there is a need for further progress to improve transparency, reproducibility, and confidence in ecology. Here, we propose a practical and operational framework to achieve good practices for building analytical procedures based on atomisation and generalisation. We introduce the concept of atomisation to identify analytical steps which support generalisation by allowing us to go beyond single analyses. These guidelines were established during the development of the Galaxy-Ecology initiative, a web platform dedicated to data analysis in ecology. Galaxy-Ecology allows us to demonstrate a way to reach higher levels of reproducibility in ecological sciences by increasing the accessibility and reusability of analytical workflows once atomised and generalised.

Linear transportation infrastructures (LTIs) are established drivers of habitat fragmentation and barrier effects. Yet, they represent an increasing surface of managed seminatural habitats where increased consideration of biodiversity outputs is needed in an era of global biodiversity decline. A combined effort by both scientists and stakeholders is, therefore, needed to evaluate the promises and limits of these alternatives so that they best achieve their conservation potential. Our study explores the effects of forest powerline clearings on biodiversity, as well as the potential benefits of integrated vegetation management (IVM) as alternatives to clear‐cuts. We recorded the acoustic activity at 35 pairs of forest/clearing stations in two forested regions of France in 2021. Our results suggest that powerline clearings represent increased movement opportunities for bats and, most particularly, edge‐foraging species. They also provide suitable habitats for bush‐cricket species, particularly species requiring thermophilic conditions. We detected no direct benefit from IVM on bat communities. However, bush‐cricket communities appeared richer, more acoustically active, and statistically different from adjacent forests in clearings favoring secondary vegetation compared with clear‐cut ones. This collaborative study provides data on understudied taxa in the context of LTIs and sheds light on conservation promises and limits associated with their management.

Many species are difficult to study either due to their rarity, elusiveness, difficult access to their area of occurrence, or any combination of these. This can be particularly problematic for threatened species. Passive acoustic monitoring (PAM) is a recently developed survey technique that has shown great potential in addressing this problem for species that communicate through vocalizations. However, the large amount of data it generates can be difficult to process manually. Here, we present an entirely automatic workflow to record and detect the vocalizations of a bird species that is both elusive (nocturnal) and restricted to difficult terrain in the most remote rainforests of an oceanic island: the recently discovered Principe Scops-Owl. Specifically, we evaluated (i) the performance of the workflow to monitor the presence of the owl, (ii) we assessed the most suitable time for monitoring it; and (iii) we examined the potential of this species to present detectable vocal individual signatures. For 12 days, we deployed omnidirectional recording stations (AudioMoth devices) in 72 points along 10 transects that were surveyed during one night at the same time by observers in the field. We trained TADARIDA, a machine learning software toolbox, to automatically detect owl calls. Results on the presence of the owl per site were similar for both methods. The automatic workflow showed that the owl is active during the whole night and the PAM recording setting should encompass at least the 21–23 h interval. Possibly, vocalizations had individual signatures—but the small sample size and temporal window prevented a definite conclusion. The automatic workflow developed here is an efficient method to monitor the Principe Scops-Owl and can be easily adapted for other elusive vocal species.

Passive acoustic monitoring can be an effective method for monitoring species, allowing the assembly of large audio datasets, removing logistical constraints in data collection and reducing anthropogenic monitoring disturbances. However, the analysis of large acoustic datasets is challenging and fully automated machine learning processes are rarely developed or implemented in ecological field studies. One of the greatest uncertainties hindering the development of these methods is spatial generalisability—can an algorithm trained on data from one place be used elsewhere? We demonstrate that heterogeneity of error across space is a problem that could go undetected using common classification accuracy metrics. Second, we develop a method to assess the extent of heterogeneity of error in a random forest classification model for six Amazonian bird species. Finally, we propose two complementary ways to reduce heterogeneity of error, by (i) accounting for it in the thresholding process and (ii) using a secondary classifier that uses contextual data. We found that using a thresholding approach that accounted for heterogeneity of precision error reduced the coefficient of variation of the precision score from a mean of 0.61 ± 0.17 (SD) to 0.41 ± 0.25 in comparison to the initial classification with threshold selection based on F-score. The use of a secondary, contextual classification with thresholding selection accounting for heterogeneity of precision reduced it further still, to 0.16 ± 0.13, and was significantly lower than the initial classification in all but one species. Mean average precision scores increased, from 0.66 ± 0.4 for the initial classification, to 0.95 ± 0.19, a significant improvement for all species. We recommend assessing—and if necessary correcting for—heterogeneity of precision error when using automated classification on acoustic data to quantify species presence as a function of an environmental, spatial or temporal predictor variable.

Passive acoustic monitoring can be an effective method for monitoring species, allowing the assembly of large audio datasets, removing logistical constraints in data collection and reducing anthropogenic monitoring disturbances. However, the analysis of large acoustic datasets is challenging and fully automated machine learning processes are rarely developed or implemented in ecological field studies. One of the greatest uncertainties hindering the development of these methods is spatial generalisability—can an algorithm trained on data from one place be used elsewhere? We demonstrate that heterogeneity of error across space is a problem that could go undetected using common classification accuracy metrics. Second, we develop a method to assess the extent of heterogeneity of error in a random forest classification model for six Amazonian bird species. Finally, we propose two complementary ways to reduce heterogeneity of error, by (i) accounting for it in the thresholding process and (ii) using a secondary classifier that uses contextual data. We found that using a thresholding approach that accounted for heterogeneity of precision error reduced the coefficient of variation of the precision score from a mean of 0.61 ± 0.17 (SD) to 0.41 ± 0.25 in comparison to the initial classification with threshold selection based on F‐score. The use of a secondary, contextual classification with thresholding selection accounting for heterogeneity of precision reduced it further still, to 0.16 ± 0.13, and was significantly lower than the initial classification in all but one species. Mean average precision scores increased, from 0.66 ± 0.4 for the initial classification, to 0.95 ± 0.19, a significant improvement for all species. We recommend assessing—and if necessary correcting for—heterogeneity of precision error when using automated classification on acoustic data to quantify species presence as a function of an environmental, spatial or temporal predictor variable.

Insects are the most diverse group of animals on Earth, but their small size and high diversity have always made them challenging to study. Recent technological advances have the potential to revolutionise insect ecology and monitoring. We describe the state of the art of four technologies (computer vision, acoustic monitoring, radar, and molecular methods), and assess their advantages, current limitations, and future potential. We discuss how these technologies can adhere to modern standards of data curation and transparency, their implications for citizen science, and their potential for integration among different monitoring programmes and technologies. We argue that they provide unprecedented possibilities for insect ecology and monitoring, but it will be important to foster international standards via collaboration.